Ranking messages of conversation graphs in a messaging platform using machine-learning signals

Abstract

According to an aspect, a method for ranking messages of conversation graphs in a messaging platform using machine-learning signals includes receiving a conversation view request to retrieve messages of a conversation graph stored on a messaging platform and determining a plurality of predictive outcomes for each of a plurality of messages of the conversation graph using predictive models. The determining step may include obtaining a plurality of signals relevant to the predictive models, the plurality of signals including data structure-related signals relating to the conversation graph and inputting the plurality of signals to the predictive models. The method includes ranking the plurality of messages based on the predictive outcomes and transmitting at least a subset of the plurality of messages to be rendered on a client application according to the rank.

What is claimed is: 
     
       1. A method for ranking messages of conversation graphs in a messaging platform using machine-learning signals, the method comprising:
 receiving, at a messaging platform, over a network, a conversation view request to retrieve messages of a conversation graph from a user device; 
 predicting, by the messaging platform, at least one engagement probability for each of a plurality of messages of the conversation graph using at least one predictive model, including: 
 obtaining a plurality of machine-learning (ML) signals relevant to the at least one predictive model, the plurality of ML signals including data structure-related signals relating to the conversation graph; and 
 inputting the plurality of ML signals to the at least one predictive model to predict a plurality of predictive outcomes including a negative engagement probability and a positive engagement probability; 
 computing an engagement value for each of the plurality of messages using the plurality of predictive outcomes; 
 ranking, by the messaging platform, the plurality of messages based on the engagement values; and 
 transmitting, by the messaging platform over the network, at least a subset of the plurality of messages to be rendered on a client application according to the rank. 
 
     
     
       2. The method of  claim 1 , wherein the conversation graph includes a tree data structure having a plurality of nodes that represent the messages of the conversation graph and edges that represent links between the nodes, the tree data structure defining one or more branches of nodes. 
     
     
       3. The method of  claim 2 , wherein the data structure-related signals include a number of replies within a branch, a number of conversations within a branch, and a number of unique authors within a branch or the conversation graph. 
     
     
       4. The method of  claim 2 , wherein the data structure-related signals include signals representing a number of the plurality of nodes and a number of the branches within the conversation graph. 
     
     
       5. The method of  claim 1 , wherein the plurality of ML signals include engagement signals representing user engagement data associated with the messages of the conversation graph, the user engagement data including engagements of users that are connected to a user of the client application in a connection graph. 
     
     
       6. The method of  claim 1 , wherein the plurality of ML signals include historical aggregate signals including engagement data associated with a user of the client application on the messaging platform. 
     
     
       7. The method of  claim 1 , wherein the plurality of ML signals include content-related signals associated with messages of the conversation graph. 
     
     
       8. The method of  claim 1 , wherein the engagement value is computed for each of the plurality of messages using the positive engagement probability and the negative engagement probability. 
     
     
       9. The method of  claim 1 , further comprising:
 obtaining, by the messaging platform, training data from a client event log that stores information received from the client application and an injection log that stores information from a timeline manager executing on the messaging platform; and 
 training, by the messaging platform, the at least one predictive model based on a machine learning algorithm inputted with the training data. 
 
     
     
       10. A system for ranking messages of conversation graphs in a messaging platform using machine-learning signals, the system including:
 a conversation graph manager configured to generate a conversation graph based on a reply structure of messages exchanged on a messaging platform, the conversation graph including a data structure of messages of a conversation; 
 a timeline manager configured to provide a stream of messages, over a network, in a social timeline of a user on a client application, the timeline manager configured to receive a conversation view request, over the network, from the client application executable by a user device; 
 an engagement predictor configured to predict, in response to the conversation view request, a plurality of predictive outcomes for each of a plurality of messages of the conversation graph using at least one predictive model, the plurality of predictive outcomes including a positive engagement probability and a negative engagement probability, 
 the engagement predictor configured to obtain a plurality of machine-learning (ML) signals relevant to the at least one predictive model, the engagement predictor configured to input the plurality of ML signals to the at least one predictive model to predict the plurality of predictive outcomes; and 
 an engagement scorer configured to compute an engagement value for each of the plurality of messages using the predictive outcomes for a respective message, 
 the timeline manager configured to rank the plurality of messages using the engagement values, and provide, over the network, at least a subset of the plurality of messages to be rendered on the social timeline according to the rank. 
 
     
     
       11. The system of  claim 10 , wherein the plurality of ML signals includes data structure-related signals relating to the conversation graph, the data structure-related signal including signals representing a number of replies within a branch, and a number of conversations within a branch between a user of the client application and an author of a root message. 
     
     
       12. The system of  claim 10 , wherein the engagement predictor is configured to obtain one or more first ML signals from a first data service executing on the messaging platform and obtain one or more second ML signals from a second data service executing on the messaging platform. 
     
     
       13. The system of  claim 10 , wherein the plurality of ML signals include at least one health-related signals, engagement signals, social graph signals, historical aggregate signals, or content-related signals. 
     
     
       14. The system of  claim 10 , wherein the plurality of predictive outcomes also include a reciprocal engagement probability, the reciprocal engagement probability including a probability value that the messaging platform is predicted to receive a reply to a respective message. 
     
     
       15. The system of  claim 10 , further comprising:
 a predictive model trainer configured to periodically train the at least one predictive model based on one or more machine learning algorithms inputted with training data, the predictive model trainer configured to provide the at least one trained predictive model to the prediction manager. 
 
     
     
       16. The system of  claim 10 , wherein the plurality of ML signals include signals generated by the client application and signals generated by the messaging platform. 
     
     
       17. The system of  claim 10 , wherein the plurality of ML signals include similarity signals, the similarity signals representing a level of similarity between a particular message and one or more messages in which a user of the client application has engaged with. 
     
     
       18. A non-transitory computer-readable medium storing executable instructions that when executed by at least one processor are configured to cause the at least one processor to:
 receive, at a messaging platform, over a network, a conversation view request for messages of a conversation graph from a user device, the conversation graph including a data structure of messages of a conversation; 
 select a candidate subset of messages from the conversation graph; 
 predict a plurality of predictive outcomes for each message of the candidate subset using at least one predictive model, including:
 obtain a plurality of machine-learning (ML) signals including data structure-related signals relating to the conversation graph; and 
 input the plurality of ML signals to the at least one predictive model to determine the plurality of predictive outcomes, the plurality of predictive outcomes including a positive engagement probability and a negative engagement probability; 
 
 compute an engagement value for each message of the candidate subset using the predictive outcomes; 
 rank the plurality of messages using the engagement values; and 
 transmit at least a subset of the plurality of messages to be rendered on a client application according to the rank. 
 
     
     
       19. The non-transitory computer-readable medium of  claim 18 , further comprising:
 obtain training data from a client event log that stores information received from the client application and an injection log that stores information from a timeline manager executing on the messaging platform; and 
 train the at least one predictive model based on a machine learning algorithm inputted with the training data. 
 
     
     
       20. The non-transitory computer-readable medium of  claim 18 , wherein the at least one predictive model includes at least one neural network.